Why do telephone keypads count from the top down, while calculators count from the bo

[wbenzoni]

I was told that they telephone keypad was inverted into order to slow down fast "dialers" (i.e. accountants and bookkeepers). Apparently there was some fear that the switches could not keep up.

All I could find was this: http://www.bellsystemmemorial.com/te...touchtone.html

[Tapioca Dextrin]

Welcome to boards. The staff report in question is unsurprisingly Why do telephone keypads count from the top down, while calculators count from the bottom up?, just keep every one on the same conference call.

[John W. Kennedy]

That link points to http://www.bellsystemmemorial.com/pdf/touchtone_hf.pdf, the actual report at the time. The adding-machine arrangement lost in the first heat.

[bibliophage]

Welcome to the SDMB, wbenzoni.

Since this is a comment on a Straight Dope Staff Report written by SDSTAFF Dex (rather than a Straight Dope column written by Cecil), I'll move this thread to the appropriate forum.

bibliophage
moderator CCC

[C K Dexter Haven]

When I did that report, lo these many years ago, I found nothing that referred to "slowing down" the speed of dialing (well, of pushing.) In the report, I said that one of the reasons for the design of the telephone keypad was

2) AT&T (the only phone company at the time) did some research that concluded there were fewer dialing errors with the 1-2-3 on top (possibly related to the traditional rotary dial layout).

So, thanks for finding the exact study, wbenzoni, interesting. And welcome to the Straight Dope Message Boards!

A side note is that the AT&T study seem to have drawn conclusions based on single studies of 20 people. Seems a pretty small sample size to me; they cite correlations later in the study, in terms of pressure, key size, letter size, etc. But I only see words like "significant" and "insignifcant" tossed about in the study on key patters, with no quantification. I always get suspicious when studies don't tell us the margin of error (or standard deviation)... and the difference in results is tenths of a second.

[Northern Piper]

Originally Posted by C K Dexter Haven

When I did that report, lo these many years ago, .....

_______________________
Last edited by SkipMagic : 07-21-2006 at 02:23 PM. Reason: Fixed coding

Man, it must suck for a Mod to have to clean up after an Admin...

[Voyager]

Originally Posted by C K Dexter Haven

A side note is that the AT&T study seem to have drawn conclusions based on single studies of 20 people. Seems a pretty small sample size to me; they cite correlations later in the study, in terms of pressure, key size, letter size, etc. But I only see words like "significant" and "insignifcant" tossed about in the study on key patters, with no quantification. I always get suspicious when studies don't tell us the margin of error (or standard deviation)... and the difference in results is tenths of a second.

The report says that 15 - 20 people were used per session, which implies to me that they used different people for each configuration. That makes sense to me. When I was at Bell Labs, long after this study, from time to time requests would to out for people to test new technology. I'm sure people at Murray Hill got asked even more often.

I suspect that your third reason for the ordering, keeping the letters in alphabetical order while keeping the letter/number association, is the answer. Notice that all the configurations that don't attempt to duplicate a dial have 1s at top. You'd think they'd try the odd two column configurations with 1 at the bottom at least once.

There are a couple of reasons that I don't believe the speed of dialing was an issue. First, I agree with Musicat that it sounds too much like the typewriter case. Second, I am almost certain that the pulses were registered at the switch. There are enough times that people don't finish dialing that making the connection at dialing speed would be very inefficient. Even back then, the registers would collect the numbers far faster than anyone can dial. Third, revenue was gotten from call length, and dialing time was not included. Anything cutting down the time the switch was in use and not making money was a good thing.

BTW, you're right on about 0 being ten pulses. At least one phone system I was on allowed you to dial by clicking the switch hook the proper number of times for each number! Interesting fact number two - the BSTJ issue giving the signaling frequencies for touch tone phones (and I think some of the codes when control was done over the voice lines) was the most checked out volume in the MIT engineering library as of 1970.

[Dead link]

This is a link I found with the filename of the dead link(and is also an image of a phone rotary dial): http://porticus.org/bell/images/dial_6aa.jpg Now I wonder if I provided this information in the right place or whether I should have sent it in to the contact link.

[Irishman]

Originally Posted by Musicat

So there is something inherent about one layout over the other that makes it possible to type faster with one? Sounds like a variation on the QWERTY/Dvorak keyboard myth.

Since the thread was revived and the question not answered...

The reason the bottom up layout would be faster for accountants and typists is basically practice. They spend hours each day for years typing numbers in the keypad layout with numbers at bottom. They get proficient with that layout. So any layout that changes the pattern so it doesn't map the same would be slower for them, because they have to stop and think.

That doesn't make the explanation true, but explains why it could be true.

[Kevbo]

I'll be darned if I can find a cite,

But I once heard that it was because accountants and such become very proficient at punching numbers extreamly quickly, and could dial faster than the touch-tones could be detected/ and or decoded by ancient cross-bar switches. The layout was thus reversed forcing them to slow down.

[John W. Kennedy]

But no one can come up with anything more than, "I heard that," or even "Somebody told me that he heard that." Whereas we actually have the original AT&T research document on-line.

[Musicat]

Originally Posted by Kevbo

I'll be darned if I can find a cite,

But I once heard that it was because accountants and such become very proficient at punching numbers extreamly quickly, and could dial faster than the touch-tones could be detected/ and or decoded by ancient cross-bar switches. The layout was thus reversed forcing them to slow down.

So there is something inherent about one layout over the other that makes it possible to type faster with one? Sounds like a variation on the QWERTY/Dvorak keyboard myth.

[John W. Kennedy]

It should be noted that the anti-Dvorak article cited is suffused with a right-wing bias, thus:

• Capitalism is the best of all possible worlds.
• QWERTY is the product of Capitalism.
• Therefore, QWERTY cannot be inferior to any other layout, and anyone who thinks another layout is better that QWERTY must be a dirty commie.

Note that I have no opinion one way or the other about the Dvorak/QWERTY question, because I lack the data. But Liebowitz and Margolis make their prejudice abundantly clear.

[Derleth]

Originally Posted by John W. Kennedy

It should be noted that the anti-Dvorak article cited is suffused with a right-wing bias

How is that relevant? The article also points up severe defecits in how Dvorak ran the experiments that "proved" his layout's superiority, and makes a strong case that when you've learned one layout the cost of switching outweighs whatever benefits might be had from using the other.

[Derleth]

It should also be noted that being pro-Capitalism is not really indicative of being right-wing.

[John W. Kennedy]

The authors blatantly start from the assumption that the Dvorak keyboard is an evil heresy. I don't trust religious fanatics whether they say they worship Allah, Jesus, or Adam Smith.

If you think that's an objective study, I've got some creationist biology textbooks to sell you.

[CurtC]

If anyone remembers the old original touch-tone phones, I can't believe that speed-dialing would be a problem with any configuration. The buttons on the old phones had strong springs and had to be mashed in pretty far to register the tone. I could do the three-finger touch-dialing technique, but it was difficult and not very fast.

[Musicat]

Originally Posted by CurtC

If anyone remembers the old original touch-tone phones, I can't believe that speed-dialing would be a problem with any configuration. The buttons on the old phones had strong springs and had to be mashed in pretty far to register the tone. I could do the three-finger touch-dialing technique, but it was difficult and not very fast.

I have a 1960 vintage TT phone right here. The pushbuttons are stiffer than a 2012 cellphone, but not too stiff to touch dial. I learned to do that in 1960 and it worked fine. Compared to a rotary, it was much faster, which is why I got it. A 1960's geek, I was.

(The buttons' tone-generating facility has deteriorated over time, and I can't reliably dial out anymore, so I just use it for incoming calls, as the audio quality is far superior to most modern phones and the handset more pleasing to hold.)

[C K Dexter Haven]

Just a mild note: until the revival yesterday, this thread is from 2006. We're not much concerned about resurrected threads, I just want to alert folks so that you don't necessarily expect responses from people who posted six years ago.

[johnpost]

the message board equivalent to "the number you have reached is no longer in service".

[qazwart]

When I worked at AT&T, I actually met Doug Kerr, one of the guys responsible for the touch tone keypad as we now know it.

There was some work done with different keyboard arrangements for the numbers. One looked at a 2 x 5 and a 5 x 2 grid, and another looked at a circular arrangement. However, what really won the day was the way the tones for the keys were generated. Bell Telephone used what was called a Dual Tone Multi-Frequency arrangement (DTMF). They could use a single transistor to generate both tones. This was important since transistors were about $10 a piece and the standard dial phone didn't cost more than $15. By arranging the buttons in a 3 x 4 arrangement, Bell Telephone only needed seven tones and they could generate them with a single transistor.

This was done back in the very late 1950s and into 1960. At that time, the old exchange names were being phased out. Your phone number was no longer GReenwood 8-1234. However, the exchange letters were still kept, so your phone number was still GR8-1234 and not 478-1234. Thus, it was still important to keep the phone dial arranged alphabetically. This meant that 1-2-3 were on the top and not 7-8-9 like on a calculator. The 0 was still on the bottom like most calculators though.

About the same time, Bell Labs was developing the first automated PBX which meant that businesses no longer needed an operator to do things like transfer a call to another number, and it was realized that it would be nice if you could have separate command keys to use these options. Thus, next to the zero, two more keys were added: A star and a diamond. This was really no problem because the DTMF arrangement meant that the two tones needed for those buttons already existed in the circuitry.

It was a long time ago, and I can't quite remember all of the details about the politics Kerr mentioned. But, it basically involved the fact that the star and diamond aren't standard characters in the ASCII character set. This meant that the user manuals couldn't be written using the typesetting software AT&T developed. Doug Kerr chose to replace them with the # and * symbols. The * was still referred to as the "star" key in the user manual, but the "#" key didn't have a name. Somehow the term "octotherp" was coined, and that became a footnote in the user manual.

Later on, Don McPherson called it the octothorpe to honor Jim Thorpe, the athlete. Since McPherson was responsible for training many of the customers, the term octothorpe became better known and was later used in external customer relations documentation.

[John W. Kennedy]

In 1965–66, my high-school computer class used to go to Murray Hill every other Saturday morning, to play with their IBM 7094 and its three slave 1460s. They had some phones there with an entire fourth column of other symbols—the only one I remember was a diamond. I believe they were used by an experimental dictation service that never went public.

[Irishman]

Originally Posted by Musicat

That doesn't match the supposed reason for the QWERTY layout, which was to force alternate hands for adjacent letters more readily than any other layout.

It wasn't supposed to match the supposed reason for the QWERTY layout. It was in response to your comment to Kevbo's comment. There is nothing inherent in the layout of 123 at top vs 123 at the bottom that makes it faster, but when you have a reasonable sized segment of the population with a lot of practice typing in one pattern such that they are extremely fast, reversing that pattern will forever throw them in a bind as habituation will drive them to be wrong.

Of course, that couldn't keep someone else from developing habituation to the new layout, or conceivably someone adapting the ability to speed type in either layout, but the principle is that people spend far less time dialing numbers than they did doing numerical entry for accounting purposes.

And again, I'm not saying it is a true reason, I am saying it is a consistent explanation.

Originally Posted by qazwart

However, what really won the day was the way the tones for the keys were generated. Bell Telephone used what was called a Dual Tone Multi-Frequency arrangement (DTMF). They could use a single transistor to generate both tones. This was important since transistors were about $10 a piece and the standard dial phone didn't cost more than $15. By arranging the buttons in a 3 x 4 arrangement, Bell Telephone only needed seven tones and they could generate them with a single transistor.

I fail to see how physical layout of the buttons affects electrical wiring underneath.

This was done back in the very late 1950s and into 1960. At that time, the old exchange names were being phased out. Your phone number was no longer GReenwood 8-1234. However, the exchange letters were still kept, so your phone number was still GR8-1234 and not 478-1234. Thus, it was still important to keep the phone dial arranged alphabetically. This meant that 1-2-3 were on the top and not 7-8-9 like on a calculator. The 0 was still on the bottom like most calculators though.

This is the more reasonable explanation. With a 3 x 4 grid, putting the letters alphabetically top down gives the standard layout. Alternately, the numbers could be arranged alphabetically starting at the bottom, but that might be less aesthetic.

[qazwart]

Originally Posted by Irishman

I fail to see how physical layout of the buttons affects electrical wiring underneath.

Each row had its own unique tone and each column had its own unique tone. Pressing a particular button selected both the row and column tones.

The tones generated were different frequencies ranges for the row and columns. Somehow, this allowed for all tones to be generated with a single transistor back in the late 1950s when the system was created. As a bonus, the dual tones somehow made the system more robust.

[Musicat]

Originally Posted by Irishman

I fail to see how physical layout of the buttons affects electrical wiring underneath.

Speaking as an engineer, I can see it well. Think of a matrix, where you have three lines vertically, one running down the 1-4-7-* column, the next under 2-5-8-0, etc. Then you have another set of lines running horizontally, the first under 1-2-3, then 4-5-6, etc. Each row generates one tone (one of the Dual Tone Multi-Frequency pair), and each column, another, single tone. One tone from a row and one tone from a column makes two tones, or DTMF.

Sure, nowdays there are other ways to do this with computers, but in 1958, hardware was far less sophisticated and far more expensive, even for simple tasks.

If you have an old TT phone (this won't work on new ones), try pressing 2 buttons in one row or one column. You will get a single tone, the one that is common to both buttons. That is one way we used to play tunes, by eliminating one of the double tones.

And to add to the excellent and interesting post by qazwart, it is not just a 3x4 matrix, but 4x4, or 16 possible tone combinations, 4 of which are not available on a standard telephone keyboard. I don't know if the ABCD ones were added later, but I suspect they were part of the original concept much like the * and # buttons were.

[friedo]

Originally Posted by Musicat

And to add to the excellent and interesting post by qazwart, it is not just a 3x4 matrix, but 4x4, or 16 possible tone combinations, 4 of which are not available on a standard telephone keyboard. I don't know if the ABCD ones were added later, but I suspect they were part of the original concept much like the * and # buttons were.

They were used by the (now defunct) military AUTOVON phone system, and they were also used for various PBX and Centrex systems for miscellaneous tasks.

Some older PBXes that are still in service will do interesting things if you feed them some 4th-column tones.

[qazwart]

Originally Posted by John W. Kennedy

In 1965–66, my high-school computer class used to go to Murray Hill every other Saturday morning, to play with their IBM 7094 and its three slave 1460s. They had some phones there with an entire fourth column of other symbols—the only one I remember was a diamond. I believe they were used by an experimental dictation service that never went public.

Those with four rows of buttons were usually sold to the military. These were the Autovon phones.

[Musicat]

Originally Posted by qazwart

Those with four rows of buttons were usually sold to the military. These were the Autovon phones.

I played with this many years ago, and you can buy DTMF generating and decoding chips for OEM applications or just tinkering. They always include 16 code pairs; it is up to the equipment designer to decide which ones to use, and for nonstandard (non-POTS) applications, what each pair represents on the receiving end.